Dissertation > Excellent graduate degree dissertation topics show

Research on the RSM-Based Multidimensional Recommendation Model

Author: LuLinZuo
Tutor: ZhouZhuRong
School: Southwestern University
Course: Applied Computer Technology
Keywords: Resource Space Model (RSM) Multidimension Recommendation Ontology Collaborative Filtering (CF)
CLC: TP391.3
Type: Master's thesis
Year: 2009
Downloads: 96
Quote: 0
Read: Download Dissertation

Abstract

Recommendation technology has become hot points in the information retrieve and fitering field with the development of Interact and E-commerce.The existing recommendation system meets the needs of people who access to information to some extent.Howerer,in many applications,it may not be sufficient to consider only users and items.There is much contextual information hidden among user’s experience.Since this contextual information does not be considered in traditional recommendation technology,it’s_unable to understand the real factors of user’s preferences to a commodity,which leads to imprecise recommendation result and lower performance.Therefore,it is an efficient improvement to incorporate relevant contextual information into the recommendation process.There are some multidimensional recommendation models which have improved the insufficiency contextual information flaw of traditional recommendation models,but there also are some problems:(1) these traditional multidimensional recommendation models were built only based on simple data cube, utilizing aggregation method similar to OLAP,which are read-only models,can’t meet the needs of frequent operations on resource in the most Internet applications.(2) They are short of semantic description,rule reasoning,dynamic updating and so on.As a semantic data model which can specify,store,manage and locate Web resources by appropriately classifying the contents of resources,RSM has aroused the domestic and foreign researcher’s interest since it was proposed.To the above problems,RSM is considered in our work.On the one hand, differing from the multidimensional data model used for data warehousing and OLAP,RSM is a semantic data model for uniformly,normally and effectively specifying and managing resources by normalizing classification semantics;On the other hand,differing from the semantic web,RSM is not a general distance space,but is based on rigorous,independent and orthogonal coordinate system,having normal forms,integrity constraints and its own algebra and calculus,which has become a complementarity in resource management by ontology.The research indicates that RSM is an effective data model for solving the multidimensional recommendation problems.Incorporating RSM into multidimensional recommendation,it not only can enhance semantic description,rule reasoning and dynamic updating,but also is of benefit to improve flexibility.Our research’s purpose is how to make users get much more useful information they need by the improved recommendation system.In the thesis,we do some research on the RSM-Based multidimensional recommendation to achieve the purpose.The research work includes the following aspects:Firstly,the characters of traditional recommendation and multidimensional recommendation are analyzed in the paper,a RSM-based multidimensional recommendation method is presented,system architecture is designed and formal description is studied.Secondly,the RSMB MRS and the semantic representation of it are defined.The MRM including MROL and operation command is established,which supports rule reasoning,dynamic updating and so on, and aggregation calculation is standardized too.Thirdly,the paper researches the construction of RSMB_MRS,which includes.①how to construct semantic resource space based on existing entity resource space is researched.Building method of video ontology,user ontology and user profile is discussed in detail.②According to the conversion method from ontology to RSM,how to construct RSMB_MRS from semantic resource space is analyzed.Finally,upon the above foundations we just mentioned,a RAA-based CF recommendation approach which incorporates reduction and aggregation algorithm into the traditional CF method is also proposed for rating estimation.The experiment using a multidimensional movie recommendation system is performed for implementing the above approach and testing its performance.The obtained results show that the RSM-based multidimensional recommendation model is an effective and feasible method in enhancing flexibility and improving semantics.Satisfaction and accruay ratio of user personal information retrieve are improved comparing with traditional OLAP based multidimensional recommendation model.Consequently,the research will improve the ability of semantic expression,the flexibility of resource operation,and the efficiency of personalized service,as well as accelerate growth of the existing recommendation technology.Moreover it has positive meaning to application research of RSM.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Retrieval machine
© 2012 www.DissertationTopic.Net Mobile